Quoting Daniel Baumann (daniel.baumann@progress-technologies.net): > On 08/01/2012 12:29 AM, Philipp Kern wrote: > > I don't think that's agreed upon. If anything default options > > should be used. > > i don't think so, see debconf bofh. > > the bottom line is that -9 compresses better than -6, and that -9 is > not a problem on amd64 and i386, see debconf talk for more information. > > open-vm-tools are only built on amd64 and i386 anyway, architectures > which are both considered to be 'fast' architectures. I don't remember Yamane-san and the whole BOF being so affirmative. The discussion that followed his initial presentation showed that "details" of the use of xz still need to be discussed, such as the defaul compression level. Hideki indeed said that more measurements should be performed on different kind of packages, in order to decide about the "right" settings. From memory, I even remember about a consensus for having a default that would be xz default....and leave it up to the maintainer to choose about using higher compression levels, depending on the size benefit they bring. So, well, as much as I am personnally convinced of the great value of a switch to xz, I don't really see any need to rush out for it (for instance, the pkg-fonts team did *not* change all our packages even though these are among those that benefit the most from it.....only Japanese fonts were switched.....and at least some weeks before the freeze, not some hours). I, for instance, ruled out switching samba to xz even if the quite high size of the package (and its prevalence in mirror downloads, see Yamane's statistics) would immediately bring a noticeable benefit. I was also somehow suspecting that release managers wouldn't like that, which the current thread seems to confirm.
Attachment:
signature.asc
Description: Digital signature